Why Not SQL: The Origin of ScopeQL
scopedb.io
scopedb.io
To try my best to avoid divergent discussion, I'd include two most significant FAQs:
*What about the interoperability with SQL?*
Some libraries and tools enable developers to write queries in a new syntax and translate them to SQL (e.g., [PRQL](https://prql-lang.org/), [SaneQL](https://www.cidrdb.org/cidr2024/papers/p48-neumann.pdf), etc.). The existing SQL ecosystem provides solid database implementations and a rich set of data tools. People always tend to think you must speak SQL; otherwise, you lose the whole ecosystem.
But wait a minute, those libraries translate their new language to SQL because they don't implement the query engine (i.e., the database) themselves, so they have to talk to SQL databases in SQL. However, ScopeQL is the query language of ScopeDB, and ScopeDB is already a database built directly on top of S3.
Thus, what we can leverage from the SQL ecosystem are data tools, such as BI tools, that generate SQL queries to implement business logic. For this purpose, one should write a translator that converts SQL queries to ScopeQL queries. Since both ScopeQL and SQL are based on relational algebra, the translation must be doable.
*Project Foo has already implemented similar features. Why not follow them?*
ScopeQL was developed from scratch but was not invented in isolation. We learn a lot from existing solutions, research, and discussions with their adopters. It includes the syntax of PRQL, SaneQL, and SQL extensions provided by other analytical databases. We also deeply empathize with the challenges outlined in the [GoogleSQL](https://research.google/pubs/sql-has-problems-we-can-fix-the...) paper.
However, as answered in the previous question, we first developed ScopeDB as a relational database. Then, we learned users' scenarios where an enhanced syntax helps maintain their business logic and increases their productivity. So, directly implementing the enhanced syntax is the most efficient way.